The concept of a tumor stem cell, a self-renewing cell capable of regenerating the tumor, has shed light on the persistence of human cancers and offers a new focus for translational research. Our goal is to understand the epigenetic mechanisms that distinguish the tumor stem cell from other cells of the tumor and from other normal, self-renewing cells. Epigenetic mechanisms, such as chromatin modifications, DNA methylation and small noncoding RNAs are stable, long-term (typically heritable) changes in the transcriptional potential of a cell that are independent of changes in the underlying genomic sequence. The epigenetic state of a cell serves to define cell identity and the limits of that cell's potential fates. Thus, knowledge of the epigenetic state of cells may identify signature for both tumor stem cell identity and potential. The epigenetic states of histone modifications, DNA methylation and small noncoding RNAs will be mapped using chromatin immunoprecipitation, bisulfite conversion and generation of RNA libraries coupled with genome-wide sequencing. For insight on tumor stem cells, we will take advantage of a previously developed model system that provides a consistent source of cells with tumor stem cell ability. Proposed experiments will examine the differences between a panel of cells, each sorted to enrich for the subpopulation with tumor stem cell ability. For comparisons, non-tumor stem cells will be isolated from the same source and the differences between tumor stem cells and pluripotent and multipotent stem cells will be examined. Additional experiments will be designed to perturb the epigenome of cells and directly test how the epigenetic state of the cell-of-origin affects subsequent tumor phenotype. The knowledge generated could lead to substantial new insights, including the identification of putative markers for early diagnosis, prognosis or monitoring of tumor therapies. Understanding the differences between tumor stem cells and other self-renewing cells could lead to more specific therapies that were also less toxic to normal cells. Manipulating the epigenomic state and examining the results on tumorigenicity would provide direct insight on how these signatures translate into clinically relevant phenotypes.